Co-Principal Investigators: Mark Lipsey (Vanderbilt University) and Mark Wilson (University of California, Berkeley)

Purpose: Statistics and data are featured prominently in the emerging national core standards in mathematics at the middle grades, but instruction in the middle grades typically isolates data from chance and fails to focus on foundations of statistical reasoning. These neglected foundations include structuring variability in data as distributions, relating data distribution to chance, and generating models of chance to guide statistical inference. Under the Data Modeling curriculum, these skills and understandings are integrated into a coordinated, cohesive curriculum. The goal of this project is to examine the efficacy of the Data Modeling curriculum on student learning and attitudes toward the mathematics of data and statistics at Grade 6. The intervention includes a sequence of 7 lessons to guide teachers in teaching key ideas, a set of coordinated formative and summative assessments items used to measure student learning and to inform instruction, and a professional development model and tools for teachers.

Project Activities: This project is intended to determine the effect of Data Modeling on student learning of and attitudes toward the mathematics of data and statistics. Forty schools will be randomly assigned to use the Data Modeling curriculum or to continue with business-as-usual. It is expected that at least two Grade 6 teachers per school will participate. Students in the experimental classrooms will be instructed in statistical reasoning using the Data Modeling curriculum. The Data Modeling lessons will be used in place of the applicable portions of their usual instruction on measurement, data analysis, statistics, and probability. These lessons are typically taught during the second semester in Grade 6 and cover 9 weeks of instruction. Teachers in the experimental classrooms will engage in professional development intended to improve their understanding of statistics while supporting their implementation of the curriculum with high fidelity.

Products: Products include evidence of the efficacy of Data Modeling on student learning and attitudes toward the mathematics of data and statistics. Peer reviewed publications will also be produced.

Structured Abstract

Setting: The study will take place in the Memphis City School district, a large urban school system.

Population: Forty schools will participate, along with 6th grade teachers and students. The students are predominantly African American (86.5 percent), from low socioeconomic status homes (86.5 percent), and are almost equal proportions of male and female students.

Intervention: Students in the experimental classrooms will be instructed in statistical reasoning using the Data Modeling curriculum. These students will investigate data display, statistics, and chance processes. Students will construct models of chance processes and use these to guide inference. Student achievement in these experimental classrooms will be contrasted to achievement in comparable classrooms where students participate in more standard curricular approaches to data and statistics. Teachers in the experimental classrooms will engage in professional development intended to improve their own understanding of statistics while supporting the implementation of the curriculum with high fidelity. Teachers will also be supported through ongoing coaching by teachers who have previously demonstrated high fidelity implementation of the curriculum.

Research Design and Methods: The research is designed as a cluster randomized control trial, with half of the 40 schools randomly assigned to the treatment condition and half to the control condition. Teachers in the treatment condition will use the Data Modeling curriculum to replace the applicable portions of their usual instruction in measurement, data analysis, statistics, and probability. Teachers will remain in the treatment condition for two years, resulting in data from two cohorts of Grade 6 students. By having the same teacher remain in the study for two years, data can be collected to examine the effects of teachers' increased experience with the intervention.

Control Condition: Students in control classes will undergo business-as-usual instruction that includes lessons on data analysis, statistics, and probability supported by one of the following texts: Holt Middle School Math, Grade 6; Everyday Counts Algebra Readiness; and EPGY Stanford Math K–7. Their teachers will not receive professional development from the Data Modeling team before or during the test years.

Key Measures: The primary student-level measures include: (a) summative and formative assessments developed under a previous IES grant that allow for granular tracing of 6 components of statistical reasoning; (b) an independently developed, distal measure of statistical literacy that will assess reasoning about data, chance, statistics, and inference; (c) mathematics subscales of the Tennessee Comprehensive Assessment Program (TCAP); and (d) a survey of student attitudes toward math. The primary treatment-level measures for assessing fidelity include teacher logs of their practices and contents of instruction, and an observational checklist of key instructional practices and processes that are integral to the data modeling approach that is completed during classroom visits by the research team.

Data Analytic Strategy: The primary analytical approach will be multilevel analyses of the student-level responses with supplementary analyses of moderator effects. A three level model will be used with students nested within classrooms nested within schools. The impact of the Data Modeling curriculum will be estimated for Cohorts 1 and 2 separately, and for both cohorts together.

Project Website: http://modelingdata.org/

Related IES Projects: Assessing Data Modeling and Statistical Reasoning (R305K060091) and Innovative Computer-Based Formative Assessment via a Development, Delivery, Scoring, and Report-Generative System (R305A120217)